The gradual loss of NPI-hood with 'need' verbs in Germanic
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
The gradual loss of NPI-hood with ‘need’ verbs in
Germanic
Germanic ‘need’ verbs exhibit a great deal of variation across time and languages with respect to
three aspects: (i) the environments in which they are licensed (strength), (ii) the array of different
syntactic patterns in which they can be used, such as transitive or raising verbs, (iii) and which of
these patterns are restricted to negative polarity licensing contexts.
One important property of ‘need’ verbs in Germanic is that they come with various syntactic
argument structures such as transitive verb with ⟨NPnom :EXP, NPacc/gen :THEME⟩, impersonal verb
⟨NPacc/gen :THEME⟩, directional phrases ⟨NPnom :EXP, NPacc/gen :THEME⟩, with non-finite control
or raising complements ⟨NPnom :EXP, VPbse/inf :THEME⟩ or finite clausal arguments ⟨NPnom :EXP,
Sthat ⟩. As demonstrated by Lightfoot (1979), Sweetser (1990), Diewald (1999) and Roberts and
Roussou (2003), the different uses of verbs with modal meaning develop at different stages in
grammaticalisation. It is well known, that circumstantial uses with infinitives developed from
transitive uses and that epistemic uses with infinitives developed from circumstantial uses:
(1) Vtrans > Vcircumstantial + INF > Vepistemic + INF.
Table 1 gives an overview over the NPI-hood of the different ‘need’-verbs in the major germanic
languages based on data from corpora (Deutsches Textarchiv, Referenzkorpus Altdeutsch, Referen-
zkorpus Mittelhochdeutsch, Nordic Dialect Corpus and Syntax Database, Wulfila Project), previ-
ous corpus studies such as Loureiro-Porto (2009) and historic dictionaries such as De Vries and
Te Winkel (1882), Verwijs and Verdam (1947), Svenska Akademiens Ordbok and Ordbog over det
danske Sprog.
As Table 1 indicates, there is an interesting correlation between the degree of grammaticalisa-
tion and the question whether a single use is distributionally unrestricted or restricted to negative
polarity environments. At the one end of the scale, there is Dutch hoeven, which is always an NPI
irrespective of the degree of grammaticalisation of the relevant uses, at the other end of the scale
there is Swedish behöva, which is only used as an NPI in is most grammaticalised use, which is the
epistemic one. All the transitive and circumstantial uses with infinitive of behöva are distribution-
ally unrestricted. In the middle of the scale there are ‘need’ verbs like German brauchen, which is
distributionally unrestricted whenever used as a transitive verb or impersonal verb, but which turns
into an NPI whenever occurring with clausal complement. In case, some use of a ‘need’ verb is
not attested it the relevant field of Table 1 shows a dash.
In a more global perspective it appears that the more grammaticalised the use of a verb is
the more likely it is to retain its negative polar status. This assumption is further corroborated
by the fact that of all the uses of English can it is only its most grammaticalised one which is
restricted to negative polar environments, as shown by Hofmann (1976: 94), Brennan (1993: 14),
Israel (1996: 630–631, 2011: 131–132), Drubig (2001: 43), Portner (2009: 30).
These data raise two questions: First of all, how do the different uses of ‘need’ verbs relate to each
other in the lexicon? Is there a single entry or are there separate and independent entries? And
secondly, why are more more grammaticalised ‘need’ verbs more likely to be NPIs?
As regards the first question, it is assumed here that lexicon entries of modal verbs in general are
organised in type hierarchies which relate all of the different uses to each other. On the top branch,
1intrans trans impers. fin. clause dir. PP inf. circ. inf. epist
Goth. þaurban + bare Inf. – NPI – NPI NPI NPI –
O. Sax. thurăan + bare Inf. – – – – NPI NPI –
O. Eng. þurfan + bare Inf. – ?NPI – – – NPI –
O. H. Ger. thurfan + bare Inf. NPI NPI – – NPI NPI –
M. H. Ger. thurfan + bare Inf. NPI NPI – – NPI NPI –
Mod. Dt. hoeven + te-Inf. – NPI NPI – NPI NPI NPI
Mod. Ger. brauchen + (zu)-Inf. – unrestr. unrestr. NPI NPI NPI NPI
Mod. Den. behøve + (at)-Inf. – unrestr. – – – NPI NPI
Mod. Eng. need + bare Inf. – unrestr. – – – NPI NPI
Mod. Nor. trenge + bare Inf. – unrestr. – – – NPI NPI
Mod. Den. behøve + bare Inf. – unrestr. – – – NPI NPI
Mod. Swe. behöva + bare Inf. – unrestr. – – – unrestr. NPI
Mod. Isl. þurfa + að-Inf. – unrestr. – – (unrestr.) unrestr. NPI
Mod. Nor. behøve + å-Inf. – unrestr. – – – unrestr. –
Mod. Nor. trenge + å-Inf. – unrestr. – – – unrestr. –
Mod. Engl. need + to-Inf. – unrestr. – – – unrestr. –
Table 1: Distribution of NPI uses of ‘need’ verbs in Germanic languages
there is the least grammaticalised use, and the more deeper in the tree, the more grammaticalised
uses tend to be (cf. 2–3)
(2) Engl. need (3) German brauchen
NP non-finite NPacc clausal
[+NPI]
non-to-inf. to-inf. trans. unpers.
[+NPI] finite non-finite
dass-clause [+NPI]
circ. [+NPI]
dir. PP bare inf.
[+NPI] [+NPI] subjbr -subjin f objbr -subjin f
[+NPI] [+NPI]
circ. epist.
[+NPI] [+NPI] dir. PP (zu) inf. bare inf.
[+NPI] [+NPI] [+NPI]
circ. epist. circ. epist.
[+NPI] [+NPI] [+NPI] [+NPI]
This internal structure of lexical polysemous verbs is motivated by evidence from language ac-
quisition, which is the main force behind grammaticalisation (cf. Paul 1920: 34 §18, Lightfoot
1979: 375, Lightfoot 1998: 18). Following Green’s (2011) concept of Type Differentiation, acqui-
sition of new forms can be understood as branching the old underspecificed form into two more
specified forms that are contrasted by conflicting feature values. Thus, grammaticalisation of new
forms can be sketched as follows: at some points of their development a f orm with the feature
F specified as a will be reanalysed. Which means it will lose its specification. In the learner’s
lexicon this is going to be expressed as the assumption of a super type f orm with a underspecified
feature F. At the same time, the L1-learner has space to assume a more grammaticalised f orm
with a feature value a which reflects a higher degree of grammaticalisation.
To illustrate this mechanism, assume the first stage of acquisition in which the transitive use of
the ‘need’ verb is acquired, consider the initial entry for German brauchen.
2⎡
⎢ NP[str] 1 , ⎤
⎥
⎢
⎢ ⎥
⎥
⎢
⎢ ⎥
⎥
(4) ⎢
⎢ SC ⟨ AGR | CASE str ⟩⎥
⎥
⎢
⎢ [ [ ]] ⎥
⎥
⎢
⎢
HEAD
noun ⎥
⎥
⎣ ⎦
Once the L1-learner is exposed to data which suggest that the THEME-argument might also be
realised as infinitive, a reanalysis takes place (i) which causes the category of the second argument
to become underspecified and (ii) which introduces two daughters one bearing the old value noun
and a second daughter bearing the new value verb, yielding a control infinitive structure. This
models exactly the data for L1 acquisition of modality and ‘need’ verbs gathered by Cournane
(2014), Cournane (2015), and Hacquard and Cournane (2016), Lin (2016) and Lin, Weerman, and
Zeijlstra (2015), Lin, Weerman, and Zeijlstra (2017).
NP[str] 1 ,
[ARG - ST ⟨ ⟩]
[HEAD noun-∨-verb]
⎡
⎢ NP[str] 1 , ⎤
⎥
⎡
⎢ NP[str] 1 , ⎤
⎥ ⎢
⎢ ⎥
⎥
⎢
⎢ ⎥
⎥ ⎢
⎢ ⎡ ⎡ ⟨ ⟩⎤⎤ ⎥
⎥
⎢ ⎥ ⎢ ⎢ ⎢ NP[str] ⎥⎥ ⎥
⎢ ⟨⎢ ⎢ ⎥⎥ ⎥
SUBJ
⎢
⎢ ⟨ ⟩⎥ ⎢ ⎢ ⎢ 1
⎥⎥⟩⎥
]] ⎥
ARG - ST AGR | CASE str ⎢ ARG - ST ⎢ ⎢ ⎥⎥ ⎥
⎢
⎢ [HEAD [ ⎥
⎥ ⎢
⎢ ⎢
⎢ HEAD ⎢
⎢ bse ⎥
⎥⎥
⎥ ⎥
⎥
⎢
⎣ noun ⎥
⎦ ⎢
⎢ ⎢
⎢ ⎢
⎢
VFORM
⎥
⎥⎥
⎥ ⎥
⎥
⎢ ⎢
⎣ ⎢
⎣ verb ⎥⎥
⎦⎦ ⎥
⎣ ⎦
(5)
Turning to the second question, there seems to be a principle at work: if any use of a ‘need’ verb
bears the NPI feature all the remaining uses which subsequently grammaticalised out of it bear
that feature too. This exactly accounts for the vast variation on NPI uses among ‘need’ verbs in
Germanic as illustrated in Table 1. In the case of Dutch, the NPI feature takes is in the top node and
inherited to all possible uses; in English the NPI feature only applies to non-finite uses without to
(cf. 2) and in German, the NPI feature extends to all the uses which involve a clausal complement
(cf. 3).
Tackling to the question why the distribution of NPI uses is so heterogenous in Modern Ger-
manic languages, it is recommended to take a look at earlier stages: as illustrated in Table 1, the
earliest documented stages Gothic, Old Saxon, Old High German and Old English involve a ‘need’
verb thurfan and its cognates which is (almost) exlusivly found in NPI licensing environments, all
above in the scope of a negation and in interrogatives, in Gothic and in Modern Swedish it appears
that relative clauses can license NPIs, too. At least it is remarkable that many of the instances
which do not occur in well known NPI-licensing contexts are found in relative clauses:
(6) sumai mundedun, ei unte arka habaida Iudas,
some-M . NOM . P mean-PRET-3 P that until/so.lange box-ACC . S have-PRET-3 S Iudas
þatei qeþi imma Iesus: bugei [REL - CL þizei
that say-OPT. PRET.3 P him-M . DAT. S Iesus: buy-IMP REL . GEN . S
þaurbeima du dulþai], aiþþau þaim unledam ei
need-OPT. PRS .1 P to feast-DAT or DEM . M . DAT. P poor- DAT. P for.that
1
hva gibau.
something-ACC give-OPT. PRS .1 S
‘For some of them thought, because Judas had the bag, that Jesus had said unto him, Buy those
things that we have need of against the feast; or, that he should give something to the poor.’
1
Wulfila Bible Codex Argenteus, John 13:29
3(7) det var ju mycket som skulle – alla lysrör [REL - CL som behövde
there was PRT a.lot that FUT. AUX all neon.lamps REL needed
2
skiftas] och andra __UNDEF__ ljuspunkter
replace-PST. PASS and other ??? light.spots
‘. . . neon lamps which needed to be replaced . . . .’
The behaviour of ‘need’ verbs in known studied stages of Germanic languages suggests thus that
the common Protogermanic ancestor *þurăan must have been negative polar covering all its uses
from transitive to clause embedding uses (cf. Birkmann 1987: 371–373 on the phonological recon-
struction).
It will be shown that almost all the ‘need’ verbs in Germanic have undergone an erosion of
their negative polarity to some extent, in some languages such as Danish the transitive uses are no
longer NPIs, in others such as Icelandic transitive and circumstantial uses lost their NPI status and
the English NPI need + bare infinitive is increasingly replaced by a non NPI need + to-infinitive
(cf. Müller 2008).
The development in the various languages indicates that both scenarios for the loss of negative
polarity can be found, replacement by a new distributionally unrestricted form (cf. Hoeksema
1998) and the loss of negative polarity (cf. Jäger 2010).
Comparing the various Modern Germanic languages, van der Wouden (2001) and Richter and
Soehn (2006) observed that ‘need’ verbs are licensed by a different types of licensing contexts in
different languages. It is argued here that the more there are NPI licensing contexts in a language,
the more difficult it is for L1 to recognise a given use as NPI in the input data, hence the more
likely it is this use is going to lose its NPI-hood. In a similar manner, Goldberg (2019: 101–104)
observes that L1-learner tend to simplify their grammars if the input becomes too opaque. All this
is in line with the well known assumption that L1-acquisition is the main locus of language change
(cf. Paul 1920: 34 §18, Lightfoot 1979: 375, Lightfoot 1998: 18). Moreover this is corroborated
by the findings on L1-acquisition of negative polar ‘need’ verbs in Lin, Weerman, and Zeijlstra
(2015) and Lin, Weerman, and Zeijlstra (2017), who show that L1-learner gradually acquire the
various licensing contexts in which Dutch hoeven ‘need’ with clausal negation niet (2;) or negative
quantifier geen ‘no’ (4;) before allowing more licensers from 7;00 onward. In other words, it takes
much time until weak NPIs are acquired. Apart from that it will be demonstrated that individual
speakers already reanalysed weak NPIs such as brauchen as distributionally unrestricted forms.
Finally, it will be shown here that there are ‘need’ verbs which are no longer strict NPIs but
which still overwhelmingly occur in non-veridical environments such as the circumstantial uses
in Norwegian trenger with infinitive (75/2) and beøve with infinitive (25/2) und to lesser extent
Swedish beöva with infinitive (66/20). These facts suggest that NPI-hood is not even a binary
feature but a gradual or probabilistic one.
Alternatively, it could be assumed that NPI-hood is not expressed by a lexical feature but a long
the lines of Israel (1996: 630–631, 2011: 127–142) who suggest that sensitivity polarity can be ex-
plained in a pragmatic way in terms of scalar implicature. As Israel points out, ‘need’-verbs encode
endpoints of a scale thereby behaving like prototypical polarity sensitive items. The account out-
lined here remains agnostic to the question whether NPI-hood is expressed as a lexical feature or
derived by pragmatic principles. But there has to be some information in the lexical entries which
designates transitive uses of need verbs to be NPIs such as in Modern Dutch, but designates them
to be distributionally unrestricted in languages like Modern Scandinavian or German.
Birkmann, Thomas (1987). Präteritopräsentia: morphologische Entwicklung einer Sonderklasse in den altgermanis-
chen Sprachen. Vol. 188. Linguistische Arbeiten. Tübingen: Niemeyer.
2
NDC: bara_om3
4Brennan, Virginia (1993). “Root and Epistemic Modal Auxiliary Verbs in English”. PhD thesis. Amherst: University
of Massachusetts.
Cournane, Ailı´is (2014). “In search of L1 Evidence for Diachronic Reanalysis”. In: Language Acquisition 21.1,
pp. 103–117.
— (2015). “Modal development: Input-Divergen L1 Acquistion in the Direction of Diachronic Reanalysis”. PhD
thesis. Toronto: University of Toronto.
De Vries, Matthias and Labert Allard Te Winkel (1882). Woordenboek der Nederlansche Taal. ’s-Gravenhage: Nijhoff.
Diewald, Gabriele (1999). Die Modalverben im Deutschen: Grammatikalisierung und Polyfunktionalität. Reihe Ger-
manistische Linguistik 208. Tübingen: Niemeyer.
Drubig, Hans Bernhard (2001). “On the syntactic form of epistemic modality”. Ms. Universität Tübingen.
Goldberg, Adele E. (2019). Explain me this. Creativity, competition, and the partial productivity of constructions.
Princeton: Princeton University Press.
Green, Georgia (2011). “Modelling grammar growth: Universal grammar without innate principles or parameters”.
In: Non-transformational syntax: a guide to current models. Ed. by Kersti Borjars and Robert Borsley. Oxford:
Wiley-Blackwell, pp. 378–403.
Hacquard, Valentine and Ailı´is Cournane (2016). “Themes and variations in the expression of modality”. In: Pro-
ceedings of NELS 46, pp. 21–42.
Hoeksema, Jack (1998). “On the (non)loss of polarity sensitivity: Duch ooit”. In: Historical linguistics 1995: Selected
th
papers from the 12 international conference on historical linguistics, Manchester, August 1995. Ed. by Richard
M. Hogg and Linda van Bergen. Amsterdam: John Benjamins, pp. 101–114.
Hofmann, Thomas Ronald (1976). “Past Tense Replacement and the Modal System”. In: Notes from the Linguistic
Underground. Ed. by James McCawley. Syntax and semantics 7. New York: Academic Press, pp. 86–99.
Israel, Michael (Dec. 1996). “Polarity sensitivity as lexical semantics”. In: Linguistics and Philosophy 19.6, pp. 619–
666. URL: https://doi.org/10.1007%2Fbf00632710.
— (2011). The grammar of polarity. Pragmatics, Sensitivity, and the Logic of Scales. Cambridge: Cambridge Uni-
versity Press.
Jäger, Agnes (2010). “Anything is nothing is something”. In: Natural Language & Linguistic Theory 28, pp. 787–822.
Lightfoot, David (1979). Principles of diachronic syntax. Vol. 23. Cambridge studies in linguistics. Cambridge: Cam-
bridge University Press.
— (1998). Development of Language: Acquisition, Change and Evolution. Maryland lectures in language and cog-
nition 1. Malden: Blackwell.
Lin, Jing (2016). “On the distribution and acquisition of the West Germanic ‘Need’ verbs: German between Dutch and
English”. In: Journal of Germanic Linguistics 28.4, pp. 349–369.
Lin, Jing, Fred Weerman, and Hedde Zeijlstra (2015). “Emerging NPIs: the acquisition of Dutch hoeven ‘need’”. In:
The Linguistic Review 32.2, pp. 333–374.
— (2017). “Acquisition of the Dutch NPI hoeven ‘need’: From lexical frames to abstract knowledge”. In: Language
Acquisition.
Loureiro-Porto, Lucía (2009). The semantic predecessors of ‘need’ in the history of English (c750–1710). Vol. 43.
Malden, MA: Wiley-Blackwell.
Müller, Friederike (2008). “From degrammaticalisation to regrammaticalisation? Current changes in the use of ‘need’”.
In: Arbeiten aus Anglistik und Amerikanistik 33.1, pp. 71–94.
Paul, Hermann (1920). Deutsche Grammatik. Vol. 4 Syntax II. Niemeyer.
Portner, Paul (2009). Modality. Oxford: Oxford University Press.
Richter, Frank and Jan-Philipp Soehn (2006). “Braucht niemanden zu scheren: A Survey of NPI Licensing in German”.
In: Proceedings of the 13th International Conference on Head-Driven Phrase Structure Grammar, Varna. Ed.
by Stefan Müller, pp. 421–440. URL: http : / / cslipublications . stanford . edu / HPSG / 2006 /
richter-soehn.pdf.
Roberts, Ian G. and Anna Roussou (2003). Synactic change: a minimalist approach to grammaticalization. Vol. 100.
Cambridge studies in linguistics. Cambridge: University Press.
Sweetser, Eve (1990). From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure. Vol. 54.
Cambridge studies in linguistics. Cambridge: University Press.
van der Wouden, Ton (2001). “Three modal verbs”. In: Zur Verbmorphologie germanischer Sprachen. Ed. by Sheila
Watts, Jonathan West, and Hans-Joachim Solms. Vol. 446. Linguistische Arbeiten. Niemeyer, pp. 189–210.
Verwijs, Eelco and Jan Verdam, eds. (1947). Middelnederlandsch Woordenboek. Nijhoff.
5You can also read
